ChatGPT replacing human investment advisors isn’t a question of ‘when’ but ‘if,’ Wharton professor says

On a recent episode of Wharton’s Ripple Effect podcast, finance professor Michael Roberts shared what he’s learned from researching AI and engaging with those who are implementing it. “It’s hard to not see [AI] as being central in one way or another to financial literacy, and financial practice, more broadly,” he said. But “where it is right now is simply nowhere near where it needs to be in terms of a stand-alone, say, investment advisor or a financial guru for individuals.”

I had a chat with Roberts to find out more, particularly his thoughts on AI’s effect on Wall Street C-suites. “AI is going to start playing an increasingly important role,” he said. “I already see it being used.” (At the Coins2Day Brainstorm AI conference in London last week, experts discussed the big banks, such as JPMorgan Chase, that were early adopters.)

Finance, fundamentally, is applied data analysis, Roberts told me, and AI is a “tool in the practitioner’s arsenal for addressing financial challenges.” In one example, he explained how students in his data science for finance course used ChatGPT to “crank out Python code, so they can focus on the financial issues rather than getting bogged down,” adding that it was “fantastic.”

However, when students put ChatGPT to the test in his applied finance course, the bot answered somewhat easy questions incorrectly. “On more complex and more subtle questions, it often struggles,” he said. “It just doesn’t know how to answer the question yet. It’s almost like a smart high schooler who takes an intuitive or naive approach to a problem.” When posed with broad, open-ended questions like, “How should I save for retirement?” Or “What should I invest in,” it didn’t know where to begin.

While Roberts not an expert in developing large language models, he is a data scientist who has been working on machine learning and AI applications in finance for over 30 years. His findings, predominantly based on using ChatGPT, show that it’s “not at a place right now where it’s going to replace investment advisors.”

“I don’t know when, or if, it will ever completely remove the need for registered investment advisors,” he added.

A recent study by researchers at Queen’s University, Virginia Tech, and JPMorgan A.I. Research that examined how large language models like ChatGPT and GPT-4 performed in mock CFA exams. And the results were underwhelming. “We concluded,” the researched wrote, “that ChatGPT would likely not be able to pass the CFA level I and II, under all tested settings.”

But in the meantime, the technology still can be quite useful when it comes to financial education—not just for those at elite business schools but, really, by anyone.

“I can’t engage with a YouTube video, but I can engage with ChatGPT,” Roberts told me. “That’s one of the reasons why it’s really going to be central to financial literacy and financial education.”

Sheryl Estrada
[email protected]

María Soledad Davila Calero curated the Leaderboard and Overheard sections of today’s newsletter.

Leaderboard

Ross Tennenbaum was promoted to president of the tax software company Avalara (formerly NYSE: AVLR). Tennenbaum has been CFO at Avalara since 2020 and will continue to serve in this role along with his new responsibilities. Previously, he worked at Credit Suisse and Goldman Sachs, where he helped Avalara go public in 2018. Avalara went back private in 2022. 

Dave Miller was named CFO at SheerID, an identity verification company. Before SheerID, Miller was head of finance and corporate controller at AWS Elemental, and VP of finance and accounting at FinancialForce. He previously held various accounting and finance leadership roles at Ernst & Young and Zynga.

Big deal

A report by Pew Research Center includes seven facts about Americans and federal taxes, drawn from the organization's surveys and analysis of federal data.

About half (53%) of Americans say the complexity of the federal tax system bothers them a lot. “Undeniably, the federal tax code is a massive document, and it has only gotten longer over time,” according to Pew. The printed 2022 edition of the Internal Revenue Code has 4,192 pages, excluding front matter. Income tax law alone accounts for 2,544 of those pages, Pew finds. 

Courtesy of Pew Research Center

Going deeper

More than 98% of the global economy’s central banks are either researching, piloting, or deploying central bank digital currency (CBDC), according to a new report by World Economic Forum and in collaboration with Accenture. The report finds that the financial sector is on the precipice of the next phase of wholesale CBDC (wCBDC) in modernizing financial markets. For example, the European Central Bank has begun testing wCBDC on distributed ledger technology for securities and foreign exchange transactions, the research finds.  

Overheard

“The ability to develop crisp mental models around the problems you want to solve and understanding 'the why' before you start working on 'the how' is an increasingly critical skill, especially in the age of AI.”

— Goldman Sachs’ chief information officer, Marco Argenti, wrote in a recent article in the Harvard Business Review. Argenti argued that those who are interested in pursuing AI engineering should also study philosophy to develop their problem solving skills.